Let's Train an AI to Play a Game
In this session, game developers from Tasmania’s Secret Lab and academics (who study everything from human-computer interaction to machine learning) will reveal what genuine AI is, how machine learning works, and how the rise of our machine overlords might still be a few decades away yet…
Live on stage, we’ll take audience suggestions and use them to train a real, actual AI to play a game. It’ll wage war, walk, navigate, or maybe just flop around the place. As you watch.
We’ll use the popular Unity game engine, and cutting edge machine learning technology (the TensorFlow framework, if you’re into this stuff). Have fun and learn a little about about AI without that pesky maths stuff! Get enthused about science, and see how games are super, super useful for this. How can a 8-limbed beast learn to walk, without animations or custom coding? How can a mighty fighting mech learn to wage an eternal war? Come find out!
Dr Paris Buttfield-Addison [Producer, Secret Lab Pty. Ltd.], Meredith Castles [Lecturer, University of Tasmania], Dr Tim Nugent [Game Developer, lonely.coffee], Dr Jon Manning [Programmer, Secret Lab Pty. Ltd.]